| | --- |
| | library_name: peft |
| | tags: |
| | - PyTorch |
| | - Transformers |
| | - trl |
| | - sft |
| | - BitsAndBytes |
| | - PEFT |
| | - QLoRA |
| | datasets: |
| | - databricks/databricks-dolly-15k |
| | base_model: meta-llama/Llama-2-7b-chat |
| | model-index: |
| | - name: llama2-7-dolly-answer |
| | results: [] |
| | license: mit |
| | language: |
| | - en |
| | --- |
| | |
| | <!-- This model card has been generated automatically according to the information the Trainer had access to. You |
| | should probably proofread and complete it, then remove this comment. --> |
| |
|
| | # llama2-7-dolly-answer |
| |
|
| | This model is a fine-tuned version of [Llama-2-7b-chat-hf](https://huggingface.co/meta-llama/Llama-2-7b-chat) on the dolly dataset. |
| | Can be used in conjunction with [LukeOLuck/llama2-7-dolly-query](https://huggingface.co/LukeOLuck/llama2-7-dolly-query) |
| |
|
| | ## Model description |
| |
|
| | A Fine-Tuned PEFT Adapter for the llama2 7b chat hf model |
| | Leverages [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://arxiv.org/abs/2205.14135), [QLoRA: Efficient Finetuning of Quantized LLMs](https://arxiv.org/abs/2305.14314), and [PEFT](https://huggingface.co/blog/peft) |
| |
|
| | ## Intended uses & limitations |
| |
|
| | Generate a safe answer based on context and a request |
| |
|
| | ## Training and evaluation data |
| |
|
| | Used SFTTrainer, [checkout the code](https://colab.research.google.com/drive/1WYlE1fTKb0WmNx0tS1hdgtcJfZ2wdOH6?usp=sharing) |
| |
|
| | ## Training procedure |
| |
|
| | [Checkout the code here](https://colab.research.google.com/drive/1WYlE1fTKb0WmNx0tS1hdgtcJfZ2wdOH6?usp=sharing) |
| |
|
| | ### Training hyperparameters |
| |
|
| | The following hyperparameters were used during training: |
| | - learning_rate: 0.0002 |
| | - train_batch_size: 32 |
| | - eval_batch_size: 8 |
| | - seed: 42 |
| | - gradient_accumulation_steps: 2 |
| | - total_train_batch_size: 64 |
| | - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
| | - lr_scheduler_type: constant |
| | - lr_scheduler_warmup_ratio: 0.03 |
| | - num_epochs: 3 |
| |
|
| | ### Training results |
| |  |
| |
|
| |
|
| | ### Framework versions |
| |
|
| | - PEFT 0.8.2 |
| | - Transformers 4.37.2 |
| | - Pytorch 2.1.0+cu121 |
| | - Datasets 2.17.1 |
| | - Tokenizers 0.15.2 |
